tony sayer wrote:
In article , Floyd L. Davidson
writes
tony sayer wrote:
Doesn't matter either way as long as its balanced working and in any
case telephone bandwidth isn't that responsive to 'ummmm...
Ahem. That is absolutely false. Telecom engineering
necessarily goes to an extreme effort to reduce what is called
"power line influence". The reasons should be obvious:
telephone and power cables are often run side by side, on the
same poles, and in the same crawl spaces, sometimes for miles at
a stretch. It is not uncommon to see as much as 40 to 50 volts
of power line AC on a telecom cable. That requires an
astounding amount of noise immunity to allow a circuit to work.
....
Yes of course you can get leakage via induction and capacitance into the
telecom lines but this does not matter as it will inevitably be induced
in both conductors and cancelled out by common mode rejection. Doesn't
matter providing the insulation in the line and transformers will stand
it to have some kilovolts actually on the line as such...
Have you ever verified what the CMRR actually is on such a
circuit? The perception that CMR just cancels out everything is
naive. Typically common mode rejection is *not* sufficient to
provide a functional dial loop on a line with 40+ volts of AC
voltage.
It varies, and CMRR may not be enough to deal with significantly
less voltage than that.
Consider that the test tone level at a customer premise telephone
set is nominally targeted at -9 dBm, and the worst case acceptable
Signal-to-Noise ratio is 24 dB, which means that all noise should be
at least at -33 dBm, which is about 0.0000005 watts. But a 40 volts
hum across a 600 ohm impedance is 2.7 watts, and there is roughly
67 dB difference!
Can you explain how your measuring or have that configured please?..
That *is* the explanation of it? All that I left out was the
arithmetic.
The maximum noise acceptable is -33 dBm (with a signal of -9 dBm
and a minimum required SNR of 24 dB). The AC voltages seen can
be in excess of 40 volts. 40 volts would be +34 dBm. That is
67 dB difference.
Common mode rejection ratios commonly are less than 65 dB on
typical cable pairs.
....
Yep but they don't use shielding on a lot of phone multicore in the UK
and it wouldn't matter anyway..
Virtually *all* "multicore" telecom cable is shielded. (Some
customer premise cable is not. But you won't find anything
within a telephone central office that isn't, and you won't find
any outside plant distribution cable that isn't.)
In a central office most all of it here is twisted pair. I think some
terminology things betwixt the UK and USA are showing up here. All the
cable co Telco multicores I've seen, though not all, are unshielded.
Virtually *all* "multicore" cable is shielded. That is *not*
individual shields on each pair, but the entire cable is inside
a (foil) sheath.
Suggesting it is otherwise is ignorant.
What do you define shielding as, just a wrap of aluminium foil with a
drain wire or a fully woven copper mesh?..
Shielding is shielding, whether it is aluminum foil or copper braid.
Where are you coming up with these ideas? Have you ever even
seen the specs for any of this?
I take it from your statements above and the lack of an answer
here that you have no experience with specifying or installing
telecommunications cabling.
Yup. I posted this URL in another message, but just in case...
here is a very interesting, if somewhat technical, article about
measured effects of grounded shielding. It is very interesting
in the context of this particular thread.
64.70.157.146/pdf/Bondingcableshields.pdf
Yes.. Thats got some good points but they don't seem to be very savvy on
some matters about EMC and RF and you can pick a few holes in that but
Heh heh, lets see you try picking any holes in it!
OK then, part 2 "On the other hand cable shields which are bonded at one
end etc". Read that thorough carefully, doesn't make sense.
"On the other hand, cable shields which are only bonded at
one end cease to provide shielding when their length exceeds
one-tenth of the wavelength of the frequencies to be
shielded against, so for example a cable 10m long only
provides any significant shielding for frequencies below
3MHz. When cable lengths exceed one-quarter of a wavelength,
shields which are bonded at one end only can become very
efficient RF antennas * radiating RF noise and picking up RF
from the environment more efficiently than if there was no
shield at all. Although the RF noise in pro-audio products
is usually caused by digital and switch-mode circuits, it
appears as common-mode (CM) noise on all the analogue inputs
and outputs too."
So be specific. It makes sense to me. What part would you like
explained?
Then take a
lump of Andrews 4-50 Heliax and see what a good radiator that is even
greater number of wavelengths . They didn't even state if it were open
circuit or terminated on a load...
Heliax is, just as they state, a good radiator if it is not
bonded properly. It provides good shielding when properly
bonded, and can become a very effective antenna at lengths
approaching or exceeding 1/4 wavelength when not bonded.
That is true regardless of whether there is a resistive load, or
not.
Please review any book on antennas! The statement made
describes the physical construction of more than one popular
variation of an antenna.
Actually we've had a lot of EMC experience over the years in radio,
audio and automotive environments and what's made by far and away the
biggest effect is bypassing of transistor junctions at RF
frequencies....
I like chocolate chip cookies myself. But that has nothing to
do with the topic we are discussing either, so I haven't brought
it up. You probably should stay on topic too?
yes their correct in screening or shielding earthing at both ends
provided that the balance in the sending and receiving ends is what it
should be, injecting current into the shield won't affect what's carried
in the encased conductors.
You didn't read it, did you? It *does* affect the signal pairs.
It reduces the noise on them, significantly.
Were is this noise coming from then?...
The paper discusses reduction of power line noise on
communications cables. Induction from nearby power wiring is
the most common source of such noise, and that is specifically
the type of noise which is reduced by allowing current flow
through the shield of a cable.
However in practice the final result is and
can be affected by transformer and electronic balanced inputs and how
"floating" they are.
In practice, what they showed was that it improves noise
immunity.
"Floating" makes no difference at all.
Well think about that, Say we have a cable the inner pairs are wrapped
around one of the power lines that you describe, and there are a LOT of
volts induced on that wiring. OK now into a transformer there will be
galvanic isolation i.e. the ends or centre tap of that transformer isn't
connected to anything.
Now take a electronically balanced input. At some
point that will be connected to say an input IC which will have supply
rails etc, and that IC will be coupled through to the output of that
line receiving amplifier now don't you think that if there were some
matter of kilovolts on said line, then that will break down the
transistor junctions ?..
It may or may not, depending on the components. But that is an
entirely different discussion. It has *nothing* to do with what
we have been talking about, and has nothing at all to do with
the paper we are currently discussion.
The point is that "floating" does not affect noise immunity.
Longitudinal balance is
the most significant factor. Magnetic shielding is ineffective
below about 10 kHz, and reverse induction via the shield (by
grounding it at both ends) is much more significant for power
line frequencies and their harmonics (which commonly exist up to
2 or 3 kHz).
I think you have that wrong. Provided that the rejection is what it
should be then whatever is induced on the pairs will cancel out.
That is simply not true. Have you ever *measured* it? It does
*not* simply cancel *everything* out.
Do you know what "longitudinal balance" is? That is the
characteristic which most determines how much is canceled out by
common mode rejection.
It is *never* perfect.
I think we could all agree that balanced working isn't really a problem.
We could all agree that common mode rejection is not always
sufficient, and that reverse induction is virtually *always*
applied to outside plant communications cables because of that.
Exactly what you mean by "balanced working", I'm not sure.
What we've been discussing. Take a signal source and connect a
transformer thereto and connect that to a pair of wires twisted together
and then connect that to another transformer and the out put winding of
that to a load. That do?..
Look up the specs on various transformers. One of those specs
will be for longitudinal balance. It is never perfect. Some
are *much* better than others. (Then look up such things a
bifilar windings, and learn more about what causes better or
worse CMRR in any given transformer design! It really is a very
interesting topic. The first thing you will note is that by
merely specifying "a transformer", you have not necessarily
provide high CMRR for your circuit...)
It is also sort of fun to play with if you never have. Set up a
hybrid bridge using transformers, and measure the isolation.
Then try getting the balance as good as you can. At one single
frequency it is possible to get as much as perhaps 70+ dB of
isolation from good transformers. But to drop that by 10-20 dB
all you have to do is put your hand on any part of the balance
circuit! Just getting near will be enough if you actually do
get a good balance.
That is all just longitudinal balance...
....
Regarding your coax circuit...
On the other hand, if you place a fluorescent light fixture close
to it, it might well hum!
Why?.
Because the shielding is not effective at powerline frequencies
and harmonics.
Regardless, that is one of the worst possible ways to wire 10
meters of cable to a microphone.
Yes agreed and you wouldn't do that, well not in a pro environment
anyway.
Now if say you ground that to the local mains earth at one end, and say
10 meters away at the microphone case end earth that to a driven rod
earth, will it or wont it hummmmmmmmmmm?.....
Your circuit is using a single ended coaxial cable. The return
path for the circuit includes the shield. Hence you've just
connected the ground differential to the signal circuit. It
won't hum if you are 100 miles from the nearest power line...
Your example is nonsense and does not demonstrate anything about
noise immunity. It merely provides and example of poor circuit
design.
--
Floyd L. Davidson http://www.apaflo.com/floyd_davidson
Ukpeagvik (Barrow, Alaska)
10 metres audio cable going into PC = too long?
..
Linear Mode
